EP2009554A1 - Method for transferring data from a source target to a destination target, and corresponding network interface - Google Patents

Method for transferring data from a source target to a destination target, and corresponding network interface Download PDF

Info

Publication number
EP2009554A1
EP2009554A1 EP20070301146 EP07301146A EP2009554A1 EP 2009554 A1 EP2009554 A1 EP 2009554A1 EP 20070301146 EP20070301146 EP 20070301146 EP 07301146 A EP07301146 A EP 07301146A EP 2009554 A1 EP2009554 A1 EP 2009554A1
Authority
EP
European Patent Office
Prior art keywords
data
address
network interface
request
target
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Ceased
Application number
EP20070301146
Other languages
German (de)
French (fr)
Inventor
Giuseppe Maruccia
Riccardo Locatelli
Lorenzo Pieralisi
Marcello Coppola
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
STMicroelectronics SA
Original Assignee
STMicroelectronics SA
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by STMicroelectronics SA filed Critical STMicroelectronics SA
Priority to EP20070301146 priority Critical patent/EP2009554A1/en
Publication of EP2009554A1 publication Critical patent/EP2009554A1/en
Application status is Ceased legal-status Critical

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F13/00Interconnection of, or transfer of information or other signals between, memories, input/output devices or central processing units
    • G06F13/14Handling requests for interconnection or transfer
    • G06F13/20Handling requests for interconnection or transfer for access to input/output bus
    • G06F13/28Handling requests for interconnection or transfer for access to input/output bus using burst mode transfer, e.g. direct memory access DMA, cycle steal

Abstract

This method for transferring data from a source target (19) to a destination target (18) in a network comprises:
- sending at least one request packet for the destination target, said request packet containing information relating to a first address where data are located, and to a second address where data are to be stored;
- sending at least one transaction request to the source target (19), said transaction request being produced from information contained in the request packet; and
- causing the source target to transfer the data located at the first address to the second address.

Description

  • The invention relates, in general, to on-chip communication architectures and is in particular directed to the transmission of data from a source target to a destination target belonging to separate interconnection systems of such on-chip communication architectures.
  • For example, the invention is thus directed to the communication of data between an on-chip bus and a so-called network-on-chip system.
  • As a matter of fact, researchers have recently proposed the network-on-chip concept (NoC) to overcome the limitations relating to the huge efforts necessary to adequately design on-chip communication systems.
  • NoC aims at providing scalable and flexible communication architectures with suitable performances. NoCs are based on a packet switched communication concept and are mainly composed of three NoCs modules, namely: a router, a network interface (NI) and a link.
  • As concerns the data format, data packets within NoC are generally composed of a header and a payload. The header is responsible for carrying all the information required for performing communication, whereas the payload contains the actual information to be transmitted.
  • Conversely, data packets transmitted over an on-chip bus are based on specific transaction protocols. For example, the so-called "ST bus" developed by the applicant is based on a "ST bus protocol type 3" using a separate request channel and a response channel to provide communication between an initiator module and a target module.
  • Conversion of data must be carried out to provide communication between such separate interconnection systems. In addition, when clocks differ, synchronization is further to be carried out to allow communication between such interconnection systems.
  • Usually, network interfaces are in particular provided to connect communicating blocks of an integrated circuit to the network and between the on-chip bus and the router of the NoC in order to convert data from one packet format to another.
  • Besides, in one approach according to the state of the art, data communication between two interconnection systems is based on the use of a processing unit associated with a direct memory access (DMA), the processing unit being used as a master unit to send a load request to a source target to retrieve data from a first address. The processing unit also sends a store request to a destination target to store the retrieved data at a destination address, previously stored in the DMA engine.
  • However, such a communication system is limited due to the fact that it is in general slow and thus very time consuming. In addition, the processing unit is unavailable during the whole data transfer process. As a matter of fact, a standard DMA engine own more than one (usually 8) channels. Each channel used for a different DMA transfer and a particular "channel x" is unavailable during a certain DMA transfer related to itself.
  • In view of the foregoing, according to one aspect, it is proposed a method for transferring data from a source target to a destination target in a network.
  • Said method comprises sending at least one request packet for the destination target, said request packet containing information relating to a first address where data are located and to a second address where data are to be stored, sending at least one transaction request from the destination target to the source target, said read request being elaborated from information contained in the request packet, and causing the source target to transfer the data located at the first address to the second address.
  • In other words, according to the state of the art, transfer of data is done through a DMA engine in order to move a certain amount of data from one memory starting from a source address to another memory starting from a destination address, the DMA engine starting transferring data to the destination target step by step.
  • Conversely, according to the proposed method for transmitting data between the source and destination targets, the network interface to which the destination target is connected sends read requests to the network interface where the source target is connected for reading information from the source target.
  • When the network interface of the destination target receives back the requested information as a reply to the read request, it stores them in the target connected to itself, starting from the destination address. Accordingly, this aspect leads to gain both in terms of performance and processing capabilities for the networks, as the network interfaces are mainly responsible for performing data communication between the source target and the destination target.
  • According to another feature, the request packet contains in addition information relating to the size of the transfer
  • The request packet may also contain information relating to the number of data transfers.
  • One field having information for identifying the method for transferring data may also be included in the request packet.
  • In such a case, said information may be coded in a header field of the request.
  • As concerns the second address, this address may be coded in the header field of the request packet.
  • Conversely, the first address may be coded in the payload of the request packet.
  • In addition, said information relating to the size of the transfer or said information relating to the number of transfer may be coded in the payload of the request packet.
  • For example, the method may comprise:
    • checking said information for identifying the method for transferring data;
    • retrieving from the header field the said second address where data are to be stored;
    • storing in memory means said second address; and
    • retrieving said first address where data are located, the number of transfers and the transfer size from the payload.
  • The method may further comprise storing in memory means of a the network interface connected to the destination target request packets specifying each at least one address where a set of data are to be retrieved.
  • For example, the transaction requests are incrementally sent from the network interface connected to the destination target to a network interface connected to the source target.
  • In addition, the location of the source target may be retrieved from a look-up table used for decoding the first address.
  • According to another aspect, it is also proposed a network interface for connecting and providing a communication in a network between a source target and a destination target, comprising means for retrieving, in at least a request packet, information relating to a first address where data to be received are located, and to a second address where data are to be stored and means for causing the source target to transfer the data located at the first address to the second address..
  • This network interface may further comprise decoding means for decoding a header of said request packet for retrieving said second address and for decoding a payload of said request for retrieving said first address.
  • According to another feature of the network interface, it may further comprise means for decoding the payload of said request packet to retrieve a size of data transfer and a number of data transfers.
  • For example, the network interface further comprises first memory means for storing transaction requests elaborated from information data retrieved from said request packet and means for incrementally sending the transaction requests to the source target.
  • It may further comprise second memory means for storing data transmitted by said request.
  • In one embodiment, the network interface comprises a first part for performing a data protocol conversion and a second part for performing protocol independent data processing.
  • For example, said first and second parts each comprise an input control block and an output control block for extracting request data in a first format from incoming request packets and storing the extracted data in said second memory means, and for reading contents from second memory means and for generating the read data signals in a second format.
  • The network interface may further comprise means for synchronizing the input and output control blocks.
  • Other advantages and features of the invention will appear on examining the detailed description of embodiments, these being in no way limiting, and of the appended drawings in which:
    • figure 1 illustrates schematically the architecture of a network interface implemented for providing communication between a network-on-chip and an on-chip bus;
    • figure 2 shows a generic architecture of the network interface of figure 1;
    • figure 3 shows the structure of a request packet header;
    • figure 4 illustrates the structure of a request packet payload;
    • figure 5 illustrates a detailed description of the network interface architecture; and
    • figure 6 illustrates the transfer of transaction requests and of responses from a distributed DMA engine to the destination target and then to the source target.
  • In the embodiment illustrated in figure 1, the network interface NI is intended to provide communication between two separate interconnect systems namely, in the illustrated example, a network-on-chip and an on-chip bus.
  • More particularly, this network interface is used to transfer data from a source target connected to a network on-chip to a destination target connected to a on-chip bus. However, it should be noted that the network interface can be used to allow communication between other types of interconnection systems and, in particular, it can be used with other on-chip protocols, namely the so-called AMBA, AXI, OCP, ...protocols.
  • As it is known, data transmitted through the network on-chip and through the on-chip bus have differing formats.
  • As a matter of fact, network on-chip is a communication architecture for systems on-chip, usually based on the concept of packet switching rather than circuit switching.
  • In on-chip bus, data are transmitted in the form of packets consisting in a set of cells that are not interruptible. In a NoC context, the packets are split into basic units called flits (Flow Control Units), and comprise a payload containing actual information to be transmitted and a header responsible for carrying all the information required for performing communication.
  • In view of the foregoing, the network interface NI is intended to provided communication between a source target connected to the network through a router and to transmit response packets to a destination target connected to the on-chip bus.
  • The network interface is thus intended to realize a conversion of request packets according to the protocol supported by the on-chip bus into request packets having the structure corresponding to the network on-chip concept. In addition, the network interface is intended to convert response packets issued from the NoC into packets corresponding to the on-chip bus.
  • In addition, the network interface is intended to allow communication between a destination target and a source target, taking into account possible clock differences.
  • Besides, although one network interface only has been illustrated, the communication system is based on a distributed Direct Memory Access concept (distributed DMA) used in systems having a unique master and several targets, in which the master acts as a DMA initiator, and is responsible for programming transfers of data between the targets connected.
  • As a matter of fact, the DMA initiator, namely a DMA engine, instructs the network interface where the destination target is connected to send to the network interface where the source target is connected a read request for a set of byte starting from an address α in the source target. When the destination target network interface receives back these bytes as a reply to the read request, it stores them in the target connected to itself, starting from the address β. Once finished, the destination target network interface propagates to the DMA engine a reply in order to signal the end of this DMA transfer.
  • In other words, the DMA initiator engine sends particular request instructions in the form of "$ load α store β" to the network interface where the destination target is connected, meaning that $ bytes starting from the address α at the source target should be transferred to the destination target from address β. This means that all communicating blocks can be either a destination target or a source target of a DMA transfer, theses targets being connected to the network on-chip with the distributed DMA network interface of figure 1, while the DMA engine initiator is connected to the network with an initiator network interface.
  • As illustrated in figure 2, the distributed DMA network interface comprises two blocks, namely a first protocol dependent module 1 and a second protocol independent module 2 having a target module and a master module.
  • The first module is responsible for the transformation required between the on-chip bus protocol and the NoC packets for normal transactions and to cope with the management of the master and the target module 2, in order to cope with the distributed DMA transfers.
  • The second module 2 is responsible for all what is a protocol independent, for example internal storage and communication with the routers.
  • As illustrated, the first module 1 communicates with the on-chip bus using bus packets corresponding to the format used in the on-chip bus, whereas the second module communicates with routers R and R' using network layer packets having, as previously indicated, a header H and a Payload.
  • To cope with the above mentioned special DMA transfer in which the DMA transfer functions are spread within the network interfaces, the initiator engine of the distributed DMA request provides the following information: the starting local address α where to store the transferred data, the starting remote address β where the data are located, and the size of the transfer. These information are coded, inside the NI initiator 15 in figure 6, within the request packet and are sent to the network interface where the destination target is connected.
  • For that purpose, the DMA request packet is tagged so that the network interface can detect the distributed DMA request. For example, as illustrated in figure 3, single bytes are used to transmit the requested information. However, it should be noted that the DMA request made by the so called "distributed DMA engine" is not tagged whereas the DMA request packet related to this DMA request and produced by the initiator NI (15 fig 6) is tagged.
  • In particular, as illustrated, the bit <23> in the request packet header is set to one by the initiator network interface connected to the distributed DMA engine.
  • Besides, the DMA engine initiator uses a specific operation code (opcode) in order to code the distributed DMA request. For example, in a specific embodiment, in the above-mentioned STBus type3 protocol a possible opcode to be used is 0 011 0110, where opcode <7>=0 and where opcode<6:4>=0 11 to indicate the size of the transaction, namely eight bytes, in this case, and opcode <3:0>=0110, to indicate, in comparison with a standard store opcode 0010, that the byte opcode <2> has been set to one and is used to signal advanced transaction.
  • Besides, header<35:28> is used to indicate the DMA initiator source and header<39:36> is used to identify the DMA transaction.
  • At last, header<71:40> is used to code the destination address for the DMA transfer.
  • The illustrated structure means that the DMA initiator gives the destination address β of the DMA transfer using the header field of the request packet.
  • Besides, the remote address where the data to be transferred are located and the size in bytes of the transfer are coded within the payload field of the request packets.
  • As indicated in figure 4, the payload field <31:0> indicates the remote address where the data are to be retrieved, the payload field <35:32> indicates the number of transfers to be carried out, and the payload field <38:36> indicates the transfer size.
  • In other words, the DMA initiator, using information coded within the request packet instructs the distributed DMA network interface connected to the destination target about the way to gather the data to be transferred from the source target. In other words, the distributed DMA engine produces a request of memory transfer, that once filtered by the initiator NI (15 fig 6) has as result the request packet related to the wanted distributed DMA.
  • As indicated further, the destination network interface stores in memory means a set of load requests, the number of which corresponds to the number of transfers. Each request is intended to retrieve a set of data from the source target, the number of bytes corresponding to the transfer size coded within the payload.
  • In addition, the DMA engine initiator may instructs the destination network interface if the load requests must be linked inside a chunk, namely a set of packets linked together by a lock signal.
  • It should be noted that the number of transfers coded within the payload field is for example coded in four bits such that it is possible to have from one to sixteen transfers relative to a single "distributed DMA" transaction. The transfer size information is instead coded for example using three bits such that the variable transfer size lies between 2 and 64 bytes (power of 2 of the value of these 3 bits).
  • Accordingly, the biggest DMA transfer available is 16 x 64, namely 1024 bytes.
  • Reference is now made to figure 5 illustrating an example of a detailed architecture of a DMA network interface.
  • As illustrated, the network interface comprises a first part, namely the first dependent protocol module 1, and a second part, namely the above mentioned independent protocol module 2 having a target module and a master module.
  • When data are transmitted from the on-chip bus through the bus interface to the NoC, the first module encodes traffic from the bus and stores it inside a First-In First-Out (FIFO) memory 4. In other words, the first module realizes a conversion of the received traffic to generate a header and a payload according to the NoC requirements and store them in the FIFO 4. The target module reads contents from the memory 4 and sends them by flits to the NoC, through the router R.
  • Besides, when data are transferred from the NoC to the bus, the second module 2 collects flits from the NoC side, realizes a packet reconstruction and puts the collected packets in a corresponding FIFO memory 5. The first module 1 thus reads the FIFO and decodes its content in order to extract the correct bus traffic.
  • It should be noted that the FIFO memories 4 and 5 are provided within the network interface in order to decouple the first module from the second module, meaning that several packets can be stored inside FIFOs by the first module before being sent. In addition, this choice leads to gain both in terms of performance and in terms of memory usage.
  • The description of the overall architecture of an example of a network interface will now be made in reference to an embodiment in which a transmission request is transmitted from a network interface connected to the network on-chip, to a source target network interface connected to the bus on-chip.
  • However, the description also applies in the case in which the request is issued from a destination target connected to the bus in order to retrieve and store data stored in a source target connected to the NoC.
  • As illustrated, the first module, the target module and the master module each comprise an input control block 6a, 6b and 6c and an output control block 7a, 7b and 7c.
  • When the request packet is received by the input control block 6a of the target module, the packets, split into flits are collected by the input control block 6a, are reconstructed, and are saved inside the FIFO memory 5 in the proper order.
  • The output control block 7b of the first module is thus responsible for taking the packet saved in the FIFO memory 5 and to produce the request for the source target according to the on-chip bus protocol.
  • It should be noted that, as the output control block 7b and the input control block 6a may operate at different frequencies, a synchronization block 8 is provided between the two blocks 6a and 7b in order to avoid metastability problems in exchanging signals used by the two blocks to correctly evaluate the full or empty status of the memory 5.
  • In addition, as already indicated, the DMA request packet can be identified by receiving distributed DMA Nis, by checking a DMA tag (figure 3) in the transport header of the request packet.
  • When the output control block 7b has checked that the header field <23> is set to one, it stores in an internal buffer some of the information coming with the request packet and stored in the FIFO memory 5, namely a DMA initiator source identifier and the local address β where the data once retrieved must be stored, and the transaction identifier given by the DMA engine initiator in order to make a distinction among multiple outstanding DMA requests.
  • In addition, from the first four bytes of the payload in the DMA request packet, the output control block 7b retrieves the remote address α of the source target containing the data which must be transferred locally to the destination target. Furthermore, from the fifth byte inside the payload, the output control block retrieves the number of transfers and the transfer size information.
  • The output control block 7b then stores in a FIFO memory 9 of the master module, N request transactions of the read type, for example N read transactions in the form "read TS", where N and TS respectively correspond to the number of transfers and to the transfer size. These read TS requests are incrementally addressed to the remote source target, starting from the remote address α. Accordingly, each read TS request has a different incremental address α+ITS in which I varies from zero to N-1.
  • It should be noted that these read transactions are mapped in the NoC packets in the same way as if they were produced by an initiator network interface connected to a master IP, namely an external master block such as a CPU or similar. It should also be noted that the network interface, and in particular the network interface of the destination target, comprises a look-up table for decoding the remote address α.
  • The read transaction TS requests extracted from the FIFO memory 9 are then converted into packets split into flits by the output control block 7c and are then forwarded to the source target.
  • N read transaction requests are emitted. An internal counter C provided for example within the output control block 7b is used to check how many read transaction requests are completed. The value of the counter initially set to N is decremented when a read transaction request is completed, each time a result of a read transaction is gathered from FIFO 10 and transformed to a write request towards the destination.
  • Each write request has thus a different incremental address in the form β+ (N-C) x TS. In other words, these are the store requests (namely N write transactions TS) emitted towards the destination target once the Read TS results have been gathered.
  • A similar counter RC initially set to 1 is also provided, for example in the input control block 6b, to check how many response packets are received in reply to the write requests. This counter RC is also decremented when such a response packed is received.
  • When the value of the counter C is zero, this means that all the read TS transactions have been propagated to the source target, read TS results are arrived in FIFO 10 and are used to produce write requests.
  • As concerns the counter RC of the input control block 6b, when the value of this counter is not zero, if the input control block receives a response from the target, it must determine that the reply is the response to the transaction. If the counter value is bigger than one, meaning that transaction requests are still pending, in this case, the RC counter is decremented and this response is skipped. If, instead, the RC value is one, the last reply of the last write transaction request related to a distributed DMA has been received. In this case, the input control block 6b stores the response coming from the target in order to produce a response for the overall memory transfer to the distributed DMA engine.
  • As illustrated in figure 5, when received, the response packet split into flits are reconstructed by the input control block 6c and are stored in a FIFO memory 10 associated with the control block 6c. This phrase is relative to the first N read TS transactions responses.
  • The thus stored packets are then retrieved by the output control block 7b of the first module for generating signals according to the format supported by the on-chip bus and delivered to the destination target.
  • A similar approach is used to transmit data from the on-chip bus to the NoC. In this case, the received packets are processed by the input control block 6b for header and payload generation and are then stored in the FIFO memory 4. The data are then retrieved by the output control block 7a for packets generation according to the format supported by the NoC and are then forwarded to the NoC.
  • At last, is should be noted that other synchronization blocks, such as 11, 12 and 13 are used to synchronize the input control block 6c and the output control block 7b and 7c, as well as the input and output control blocks 6b and 7a.
  • Referring now to figure 6, an example of the path of the request DMA packets from a DMA engine 14 to a proper DMA network interface is given.
  • As previously indicated, the DMA request issued by the distributed DMA engine (arrow A1) are transmitted to the DMA engine network 15 interface that produces a DMA request packet that is then delivered to the proper network interface of the destination target (arrow A2).
  • As also previously indicated, incoming data transmitted by the request are retrieved and stored within the FIFO memory of the network interface and then decoded by the output control block of the corresponding first module.
  • The N request transaction requests of the type "read TS" are then stored in the FIFO memory 9 of the corresponding master module (arrows A3 and A4) and then transmitted to the source target NI 17 (arrow A5) and to the target 19 (arrows A6 and A7). Data are then retrieved by the network interface 17 connected to the network and, in particular, by the input control block 6b, to be transmitted, through the output control block 7a of the target module 2, to the input control block 6c of the master module of the network interface connected to the destination target (arrows A8, A9 and A10). After packet reconstruction and storage of the packets in the FIFO memory 10 (figure 5), the retrieved data are transmitted to the destination target 18 using N write TS transactions (arrows A11 and A12).
  • Thereafter, N-1 responses to the first N-1 write TS transactions are transmitted to the target module (arrows A13 and A14). The last response, namely the Nth one, is the one that is used as a response for the DMA request and reaches the DMA initiator 14 (arrows A15 and A16).

Claims (21)

  1. Method for transferring data from a source target (16) to a destination target (18) in a network, characterized in that the method comprises:
    - sending at least one request packet for the destination target, said request packet containing information relating to a first address where data are located, and to a second address where data are to be stored;
    - sending at least one transaction request from the destination target (18) to the source target (19), said read request being elaborated from information contained in the request packet; and
    - causing the source target to transfer the data located at the first address to the second address.
  2. Method according to claim 1, wherein the request packet contains, in addition, information relating to the size of the transfer.
  3. Method according to claims 1 or 2, wherein the request packet contains in addition information relating to the number of data transfers.
  4. Method according to any of claims 1 to 3, wherein the request packet contains one field having information for identifying the method for transferring data.
  5. Method according to claim 4, wherein said information is coded in a header field (H) of the request packet.
  6. Method according to claim 5, wherein the second address is coded in the header field of the request packet.
  7. Method according to any of claims 1 to 5, wherein the first address is coded in a payload of the request packet.
  8. Method according to any of claims 2 to 7, wherein said information relating to the size of the transfer is coded in a payload of the request packet.
  9. Method according to any of claims 3 to 8, wherein the number of transfers is coded in the payload of the request packet.
  10. Method according to any of claims 5 to 9, further comprising:
    - checking said information for identifying the method for transferring data;
    - retrieving from the header field (H) said second address where data are to be stored;
    - storing in memory means said second address; and
    - retrieving said first address where data are located, the number of transfers and the transfer size from the payload.
  11. Method according to claim 10, further comprising storing in memory means (9) of a network interface connected to the destination target request packets specifying each at least one address where a set of data are to be retrieved.
  12. Method according to claim 11, further comprising incrementally sending the transaction request from the network interface connected to the destination target to a network interface connected to the source target.
  13. Method according to any of claims 1 to 12, further comprising retrieving the location of the source target (19) from a look-up table for decoding the first address.
  14. Network interface for connecting and providing communication in a network from a source target to a destination target, comprising means (7b) for retrieving in at least a request packet information relating to a first address where data to be received are located, and to a second address where data are to be stored, and means (9, 7c) for causing the source target to transfer the data located at the first address to the second address.
  15. Network interface according to claim 14, comprising decoding means (7b) for decoding a header (H) of said request packet for retrieving said second address and for decoding a payload of said request for retrieving said first address.
  16. Network interface according to claim 15, further comprising means for decoding the payload of said request packet to retrieve a size of data transfer and a number of data transfers.
  17. Network interface according to any of claims 14 to 15, further comprising first memory means (9) for storing transaction requests elaborated from information data retrieved from said request packet and means (7b) for incrementally sending the transaction request packet to the source target.
  18. Network interface according to claim 17, further comprising second memory means (7b) for storing data transmitted by said request.
  19. Network interface according to any of claims 14 to 18, further comprising a first part (1) for performing data protocol conversion and a second part (2) for performing protocol independent data processing.
  20. Network interface according to claim 19, wherein said first and second part each comprise an input control block (6a, 6b, 6c) and an output control block (7a, 7b, 7c) for extracting request data in a first format from incoming request packets and storing the extracted data in the second memory means and for reading contents from the second memory means and generating the read data signals in a second format.
  21. Network interface according to claim 20, further comprising means for synchronising the input and output control blocks.
EP20070301146 2007-06-25 2007-06-25 Method for transferring data from a source target to a destination target, and corresponding network interface Ceased EP2009554A1 (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
EP20070301146 EP2009554A1 (en) 2007-06-25 2007-06-25 Method for transferring data from a source target to a destination target, and corresponding network interface

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
EP20070301146 EP2009554A1 (en) 2007-06-25 2007-06-25 Method for transferring data from a source target to a destination target, and corresponding network interface
US12/143,196 US8352628B2 (en) 2007-06-25 2008-06-20 Method for transferring data from a source target to a destination target, and corresponding network interface

Publications (1)

Publication Number Publication Date
EP2009554A1 true EP2009554A1 (en) 2008-12-31

Family

ID=38683574

Family Applications (1)

Application Number Title Priority Date Filing Date
EP20070301146 Ceased EP2009554A1 (en) 2007-06-25 2007-06-25 Method for transferring data from a source target to a destination target, and corresponding network interface

Country Status (2)

Country Link
US (1) US8352628B2 (en)
EP (1) EP2009554A1 (en)

Families Citing this family (17)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR100653087B1 (en) * 2005-10-17 2006-11-25 삼성전자주식회사 System for axi compatible network on chip and interleaving method thereof
WO2008038235A2 (en) * 2006-09-27 2008-04-03 Ecole Polytechnique Federale De Lausanne (Epfl) Method to manage the load of peripheral elements within a multicore system
US9465771B2 (en) 2009-09-24 2016-10-11 Iii Holdings 2, Llc Server on a chip and node cards comprising one or more of same
US9054990B2 (en) 2009-10-30 2015-06-09 Iii Holdings 2, Llc System and method for data center security enhancements leveraging server SOCs or server fabrics
US9680770B2 (en) * 2009-10-30 2017-06-13 Iii Holdings 2, Llc System and method for using a multi-protocol fabric module across a distributed server interconnect fabric
US8599863B2 (en) * 2009-10-30 2013-12-03 Calxeda, Inc. System and method for using a multi-protocol fabric module across a distributed server interconnect fabric
US9876735B2 (en) 2009-10-30 2018-01-23 Iii Holdings 2, Llc Performance and power optimized computer system architectures and methods leveraging power optimized tree fabric interconnect
US9077654B2 (en) 2009-10-30 2015-07-07 Iii Holdings 2, Llc System and method for data center security enhancements leveraging managed server SOCs
US20110103391A1 (en) 2009-10-30 2011-05-05 Smooth-Stone, Inc. C/O Barry Evans System and method for high-performance, low-power data center interconnect fabric
US9311269B2 (en) 2009-10-30 2016-04-12 Iii Holdings 2, Llc Network proxy for high-performance, low-power data center interconnect fabric
JP2012146201A (en) 2011-01-13 2012-08-02 Toshiba Corp On-chip router and multi-core system using the same
US20130107444A1 (en) 2011-10-28 2013-05-02 Calxeda, Inc. System and method for flexible storage and networking provisioning in large scalable processor installations
US9092594B2 (en) 2011-10-31 2015-07-28 Iii Holdings 2, Llc Node card management in a modular and large scalable server system
JP2013196167A (en) * 2012-03-16 2013-09-30 Toshiba Corp Information processor
JP5624579B2 (en) * 2012-03-23 2014-11-12 株式会社東芝 On-chip router
US9648102B1 (en) 2012-12-27 2017-05-09 Iii Holdings 2, Llc Memcached server functionality in a cluster of data processing nodes
US20150012679A1 (en) * 2013-07-03 2015-01-08 Iii Holdings 2, Llc Implementing remote transaction functionalities between data processing nodes of a switched interconnect fabric

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6081851A (en) * 1997-12-15 2000-06-27 Intel Corporation Method and apparatus for programming a remote DMA engine residing on a first bus from a destination residing on a second bus
US20050238035A1 (en) * 2004-04-27 2005-10-27 Hewlett-Packard System and method for remote direct memory access over a network switch fabric
US20060075057A1 (en) * 2004-08-30 2006-04-06 International Business Machines Corporation Remote direct memory access system and method

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6341318B1 (en) * 1999-08-10 2002-01-22 Chameleon Systems, Inc. DMA data streaming
US7508846B2 (en) * 2002-11-22 2009-03-24 Nortel Networks Ltd. Physical capacity aggregation system and method
US20070041383A1 (en) * 2005-04-05 2007-02-22 Mohmmad Banikazemi Third party node initiated remote direct memory access
KR100687659B1 (en) * 2005-12-22 2007-02-21 삼성전자주식회사 Network interface of controlling lock operation in accordance with axi protocol, packet data communication on-chip interconnect system of including the network interface, and method of operating the network interface

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6081851A (en) * 1997-12-15 2000-06-27 Intel Corporation Method and apparatus for programming a remote DMA engine residing on a first bus from a destination residing on a second bus
US20050238035A1 (en) * 2004-04-27 2005-10-27 Hewlett-Packard System and method for remote direct memory access over a network switch fabric
US20060075057A1 (en) * 2004-08-30 2006-04-06 International Business Machines Corporation Remote direct memory access system and method

Also Published As

Publication number Publication date
US8352628B2 (en) 2013-01-08
US20080320161A1 (en) 2008-12-25

Similar Documents

Publication Publication Date Title
US7287113B2 (en) Method of and apparatus for controlling bidirectional streams of isochronous data flowing between an application and a bus structure
EP1358562B1 (en) Method and apparatus for controlling flow of data between data processing systems via a memory
US6414966B1 (en) Bridging device for mapping/demapping ethernet packet data directly onto and from a sonet network
US5959994A (en) ATM/SONET network enhanced as a universal computer system interconnect
US5745684A (en) Apparatus and method for providing a generic interface between a host system and an asynchronous transfer mode core functional block
US7639701B2 (en) Packet format for a distributed system
US4484326A (en) Packet load monitoring by trunk controllers
JP3783731B2 (en) Fast switching network structure
US7206283B2 (en) High-performance network switch
US7978702B2 (en) Backplane interface adapter
US6513085B1 (en) Link/transaction layer controller with integral microcontroller emulation
CN103229155B (en) High-speed memory system
US4536873A (en) Data transmission system
US5610745A (en) Method and apparatus for tracking buffer availability
US5751951A (en) Network interface
US20030099254A1 (en) Systems and methods for interfacing asynchronous and non-asynchronous data media
US4993025A (en) High efficiency image data transfer network
US20100027559A1 (en) Transmission device and data extended transmission method
US9094237B2 (en) Packet routing and switching device
US20040179548A1 (en) Method and system for encoding wide striped cells
US6625671B1 (en) Compression of buffered data
US20020091884A1 (en) Method and system for translating data formats
EP0709985A2 (en) Path allocation system and method for a high performance fiber optic switch
US5400326A (en) Network bridge
US7609718B2 (en) Packet data service over hyper transport link(s)

Legal Events

Date Code Title Description
AK Designated contracting states:

Kind code of ref document: A1

Designated state(s): AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HU IE IS IT LI LT LU LV MC MT NL PL PT RO SE SI SK TR

AX Extension or validation of the european patent to

Countries concerned: ALBAHRMKRS

17P Request for examination filed

Effective date: 20090618

17Q First examination report

Effective date: 20090721

AKX Payment of designation fees

Designated state(s): DE FR GB IT

18R Refused

Effective date: 20110525